Revert to 0.14.0, remove internal strategy docs, honest README status, fix ADK tools fail-open

.agents/skills/agents-shipgate/assets/advisory-pr-comment.yml

@@ -1,3 +1,3 @@
 # Advisory PR comment.
 # Recommended starting point: runs the scanner on every PR, posts a summary
 # comment, uploads the report as an artifact, and never fails the job.
@@ -18,9 +18,9 @@
       - uses: actions/checkout@v4
         with:
           fetch-depth: 0
-      - uses: ThreeMoonsLab/agents-shipgate@v1.0.0a1
+      - uses: ThreeMoonsLab/agents-shipgate@v0.14.0
         with:
           ci_mode: advisory
           diff_base: target
           pr_comment: 'true'
-          shipgate_version: '1.0.0a1'
+          shipgate_version: '0.14.0'

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -11,7 +11,7 @@ body:
     id: version
     attributes:
       label: Agents Shipgate version
-      placeholder: "v1.0.0a1"
+      placeholder: "v0.14.0"
     validations:
       required: true
   - type: dropdown

.gitignore

@@ -25,28 +25,7 @@ harness/adoption/artifacts/
 .claude/*
 !.claude/commands/
 
-# Launch/deck workspaces
-docs/decks/**/scratch/
-docs/decks/**/.DS_Store
-docs/decks/**/.~lock*
-docs/decks/**/~$*
-docs/decks/**/*.tmp
-docs/decks/**/qa*.jpg
-docs/decks/vc-thesis/slide-08-options/*.html
-docs/decks/vc-thesis/slide-08-options/_v3_*.png
 samples/**/.agents-shipgate/baseline.json
 
-# Keep curated final VC deck artifacts despite the generic build/ ignore.
-!docs/decks/vc-thesis/build/
-docs/decks/vc-thesis/build/*
-!docs/decks/vc-thesis/build/fragments/
-docs/decks/vc-thesis/build/fragments/*
-!docs/decks/vc-thesis/build/_logo-mark-*.png
-!docs/decks/vc-thesis/build/contact-sheet*.png
-!docs/decks/vc-thesis/build/deck*.pdf
-!docs/decks/vc-thesis/build/deck*.pptx
-!docs/decks/vc-thesis/build/slide-*.png
-!docs/decks/vc-thesis/build/fragments/*.png
-
 # Merged-PR miner clone cache (benchmark/miner)
 .miner-work/

.well-known/agents-shipgate.json

@@ -3,7 +3,7 @@
   "name": "agents-shipgate",
   "display_name": "Agents Shipgate",
   "tagline": "The deterministic merge gate for AI-generated agent capability changes",
-  "version": "1.0.0a1",
+  "version": "0.14.0",
   "license": "Apache-2.0",
   "publisher": {
     "name": "Three Moons Lab",
@@ -58,12 +58,12 @@
   ],
   "package": {
     "pypi": "agents-shipgate",
-    "github_action": "ThreeMoonsLab/agents-shipgate@v1.0.0a1",
+    "github_action": "ThreeMoonsLab/agents-shipgate@v0.14.0",
     "github_repo": "ThreeMoonsLab/agents-shipgate"
   },
   "release_status": {
     "track": "verify-capable release",
-    "latest_release": "v1.0.0a1"
+    "latest_release": "v0.14.0"
   },
   "install": {
     "pipx": "pipx install agents-shipgate",

CHANGELOG.md

@@ -2,6 +2,18 @@
 
 ## Unreleased
 
+## 0.14.0 - 2026-06-30
+
+- **Versioning: the `1.0.0-alpha` line is withdrawn; this work ships as
+  `0.14.0`.** An earlier draft of this cycle briefly carried `1.0.0a1`. That
+  label was withdrawn: the `report.json` schema (`report_schema_version:
+  "0.28"`) is still additive-versioned and not yet frozen, the package is still
+  `Development Status :: 4 - Beta`, and no real-world detection-accuracy
+  baseline has been published — none of which support a `1.0` line. `0.14.0`
+  continues the `0.x` contract line from `0.13.0` and carries the same
+  agent-controller cleanup (see
+  [STABILITY.md](STABILITY.md#migration-note-0-14-0)). A `1.0` line will begin
+  only when the report schema reaches `1.0` and holds without a breaking change.
 - **Non-preview `verify` now fails closed on a missing `--config`.**
   `agents-shipgate verify --workspace . --config missing.yaml --json` exits
   `2` with `merge_verdict: "unknown"`, `applicability: "unknown"`, and

CONTRIBUTING.md

@@ -42,10 +42,9 @@ agents-shipgate list-checks
 ## Surface discipline
 
 Read this before adding a new public surface. This project has shipped surface
-area faster than it has proven the surface it already has — the review in
-[`docs/shipgate-strategic-engineering-review.md`](docs/shipgate-strategic-engineering-review.md)
-names this directly. Until the verdict-accuracy benchmark and default-on
-activation land, the bar for new surface is deliberately high.
+area faster than it has proven the surface it already has. Until the
+verdict-accuracy benchmark and default-on activation land, the bar for new
+surface is deliberately high.
 
 A **new surface** is any of: a new CLI command or sub-app; a new
 `report_schema_version` or other versioned schema; a new top-level report or

README.md

@@ -21,6 +21,15 @@ Local-first and static by default — no agent execution, tool calls, LLM calls,
 
 <!-- Canonical tagline: The deterministic merge gate for AI-generated agent capability changes. -->
 
+> [!IMPORTANT]
+> **Status: pre-1.0 (beta).** The decision engine is deterministic and stable,
+> but Shipgate's real-world detection accuracy is still being validated against
+> a labeled corpus of agent PRs — no precision/recall numbers are published yet.
+> On heavily dynamic tool surfaces (factory-built toolsets, config-bound
+> allowlists, runtime-assembled tools), Shipgate deliberately returns
+> `insufficient_evidence` rather than guess. Treat it as an advisory gate while
+> that accuracy work is in progress — see [ROADMAP.md](ROADMAP.md).
+
 ## 60 seconds: watch it block two PRs
 
 Claude Code adds `stripe.create_refund` to your support agent and opens a
@@ -45,6 +54,15 @@ uvx agents-shipgate fixture run agent_weakens_gate
 gate-removal checks are suppression-immune: the cheapest reward-hack is
 also the most visible one.
 
+**…and here's the failure mode.** These two cases are constructed fixtures with
+a clear-cut answer, chosen to show the gate working. Real PRs are messier: when
+a change builds its tool surface dynamically — a toolkit factory, a config-bound
+allowlist, tools assembled at runtime — static extraction often can't enumerate
+the result, and Shipgate returns `insufficient_evidence` and routes to a human
+rather than emit a confident wrong verdict. That is the intended failure mode,
+not a bug; reducing how often it fires on real dynamic code is active work (see
+[ROADMAP.md](ROADMAP.md)).
+
 One engine decides (`report.json.release_decision.decision`); everything
 else — `merge_verdict`, PR comments, Check Runs, Action outputs — is a
 deterministic projection of it. Five-minute version:
@@ -403,7 +421,7 @@ jobs:
       - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd
         with:
           fetch-depth: 0
-      - uses: ThreeMoonsLab/agents-shipgate@v1.0.0a1
+      - uses: ThreeMoonsLab/agents-shipgate@v0.14.0
         with:
           ci_mode: advisory
           diff_base: target

ROADMAP.md

@@ -2,7 +2,7 @@
 
 > **Naming.** This project is **Agents Shipgate** (display name) / `agents-shipgate` (package, CLI, repo). See [`AGENTS.md` § Naming (canonical)](AGENTS.md#naming-canonical) for the full convention.
 
-**Latest release: `v1.0.0a1`** — the **agent-native contract cleanup** cycle.
+**Latest release: `v0.14.0`** — the **agent-native contract cleanup** cycle.
 
 ## What Agents Shipgate is
 

STABILITY.md

@@ -1,20 +1,31 @@
-# Stability Contract · 1.0.0-alpha
+# Stability Contract · 0.14.0
 
 What agents and CI integrations can rely on across versions of Agents Shipgate.
 
 This document is the contract. If the runtime ever diverges from what's documented here, that's a bug — please file an issue.
 
+Shipgate is pre-1.0. The CLI surface, exit codes, and `contract_version`
+described here are stable within the `0.x` line, but the `report.json` schema
+(`report_schema_version`, currently `0.28`) is still additive-versioned and
+not yet frozen. A `1.0` line will not begin until the report schema reaches
+`1.0` and holds without a breaking change. Pin a version (or the Action tag)
+for reproducible CI.
+
 ---
 
-<a id="migration-note-100-alpha"></a>
+<a id="migration-note-0-14-0"></a>
 
-## Migration Note 1.0.0 Alpha
+## Migration Note: 0.14.0
 
-`1.0.0a1` starts a new alpha contract line on top of the `0.13.0`
-release. It deliberately cleans up overlapping agent-controller contracts
-instead of preserving every `0.x` surface.
+`0.14.0` continues the `0.x` contract line from `0.13.0`. It is a minor
+release that nonetheless makes deliberate breaking changes to the
+agent-controller surface — permitted under `0.x` semantics — cleaning up
+overlapping contracts instead of preserving every earlier surface. (An
+earlier draft of this work was briefly labelled `1.0.0-alpha`; that label was
+withdrawn because the report schema is not yet frozen, and the same changes
+ship here as `0.14.0`.)
 
-Breaking changes from the `0.x` line:
+Breaking changes from the `0.13.0` line:
 
 - `agents-shipgate verify` no longer writes
   `agents-shipgate-reports/agent-result.json`. Agents should read
@@ -67,12 +78,12 @@ Breaking changes from the `0.x` line:
 `verifier.json.merge_verdict` is the controller projection for agents and
 PR automation; it is not a second release gate.
 
-## What WILL NOT change in the current alpha line
+## What WILL NOT change in the current `0.x` line
 
 ### CLI command surface
 
-These commands and flags are stable across the current `1.0.0a*`
-contract line. Future alpha versions may make deliberate breaking
+These commands and flags are stable across the current `0.14.x`
+contract line. Future `0.x` versions may make deliberate breaking
 changes only by bumping `contract_version` and updating this file.
 
 | Command | Stable flags |
@@ -108,7 +119,7 @@ changes only by bumping `contract_version` and updating this file.
 ### Provisional CLI command surface
 
 The org/fleet governance commands are preview surfaces in the current
-`1.0.0a*` line. They are documented, deterministic, local-only, and included in
+`0.14.x` line. They are documented, deterministic, local-only, and included in
 `agents-shipgate contract --json` / `.well-known/agents-shipgate.json` for
 design-partner discovery, but their flags and schemas are not stable
 command-contract commitments yet. They remain consumers of `verify` artifacts;

adoption-kits/claude-code-skill/SKILL.md

@@ -1,3 +1,3 @@
 ---
 name: agents-shipgate
 description: Run prominent Agents Shipgate flows when a change touches what an AI agent can do: `shipgate check`, `agents-shipgate verify`, or `shipgate audit --host`. Use after adding or modifying MCP servers or tools, tool/function definitions (@tool, @function_tool), OpenAPI specs that describe agent tools, agent prompts, permission scopes, approval or confirmation policies, agent CI workflows, or shipgate.yaml — and before creating a PR for any such change. Also use to verify agent-related PRs, fix or triage Shipgate findings, add Shipgate to CI, or interpret Shipgate verifier/report artifacts. Triggers on phrases like "add shipgate", "verify this agent PR", "merge verdict", "release readiness for my agent", "tool-use readiness", "shipgate check", "agents-shipgate verify", "audit host grants", "shipgate.yaml", "agents-shipgate-reports/verifier.json", "agents-shipgate-reports/report.json", "fix shipgate finding".
@@ -72,7 +72,7 @@
 
 ## Stable contracts (rely on these)
 
-- **CLI surface** follows the current alpha contract line — see https://github.com/ThreeMoonsLab/agents-shipgate/blob/main/STABILITY.md.
+- **CLI surface** follows the current 0.x contract line — see https://github.com/ThreeMoonsLab/agents-shipgate/blob/main/STABILITY.md.
 - **Installed CLI contract**: when available, run `agents-shipgate contract --json` to verify local schema versions, capability/research surfaces, `release_decision.decision`, and manual-review signal fields. Older installs should use [`docs/agent-contract-current.md`](https://github.com/ThreeMoonsLab/agents-shipgate/blob/main/docs/agent-contract-current.md) or upgrade before automating against the local contract command.
 - **Verifier JSON**: `verifier_schema_version: "0.1"`. Read `merge_verdict`, `can_merge_without_human`, `first_next_action`, `fix_task`, `capability_review.top_changes`, `trust_root_touched`, and `policy_weakened` before summarizing an AI-generated PR. `merge_verdict` is a deterministic projection; the gate remains `report.json.release_decision.decision`.
 - **Verify run JSON**: `verify-run.json` uses `schema_version: "shipgate.verify_run/v1"` and records stable run identity, subject refs, input hashes, outcome, and artifact hashes. It is the reproducibility artifact for `verify`; do not treat it as a second gate.

adoption-kits/claude-code-skill/prompts/add-shipgate-to-repo.md

@@ -1,3 +1,3 @@
 # Prompt · Add Agents Shipgate to a repo
 
 You are working in a repo that may contain an AI agent — likely one of: an MCP server tool list (`*mcp*.json` or `.agents-shipgate/*.json`), an OpenAPI spec the agent calls, a Codex plugin package (`.codex-plugin/plugin.json`) or marketplace (`.agents/plugins/marketplace.json`), a Python file with `@function_tool` / `@tool` decorators (OpenAI Agents SDK, LangChain, CrewAI), a Google ADK agent in `agent.py`, an Anthropic Messages API artifact set under `prompts/`/`tools/anthropic-tools.json`/`policies/anthropic-policy.yaml`, or an OpenAI API artifact set under `prompts/`/`tools/openai-tools.json`/`openai-config.json`.
@@ -11,16 +11,16 @@
 
 1. **Install the tool - pin the version so a stale build can't shadow it.** This flow uses the current verifier, agent-handoff, primary-command, and Codex-boundary contracts and requires **contract v9 or newer**; an older copy lingering on `PATH` may lack the command or schema fields this prompt expects. Prefer a **pinned, zero-install** runner that fetches the exact version every time instead of trusting whatever is already on `PATH`. **Pin it into one variable and use that for every step below**, so no single command can fall through to a stale binary:
    ```bash
-   SG="uvx agents-shipgate@1.0.0a1"           # uv: ephemeral, always the pinned build
-   # or: SG="pipx run agents-shipgate==1.0.0a1"
+   SG="uvx agents-shipgate@0.14.0"           # uv: ephemeral, always the pinned build
+   # or: SG="pipx run agents-shipgate==0.14.0"
    $SG --version                             # confirm the pinned runner resolves
    ```
    Every step below calls `$SG …`; e.g. `$SG verify --preview --json` runs the verify preview through the pinned runner, never a `PATH` copy.
 
    If you would rather install onto `PATH`, pin the floor and **fail loudly when it resolves older** — a plain `pipx install agents-shipgate` is a no-op when an older build already exists — then set `SG=agents-shipgate`:
    ```bash
    python -m pip install -U --pre agents-shipgate
-   agents-shipgate contract --json   # STOP if this reports contract_version < 9 - re-run pinned via uvx agents-shipgate@1.0.0a1
+   agents-shipgate contract --json   # STOP if this reports contract_version < 9 - re-run pinned via uvx agents-shipgate@0.14.0
    SG=agents-shipgate                # only after the line above confirms contract v9+
    ```
 

adoption-kits/claude-code-skill/prompts/decide-shipgate-relevance.md

@@ -1,3 +1,3 @@
 # Prompt · Decide whether Agents Shipgate is relevant
 
 You are working in a repo or reviewing a PR and need to decide whether
@@ -71,13 +71,13 @@
 5. **Then act.**
    - If `run_shipgate: true`: run verify preview through a **version-pinned,
      zero-install** runner so a stale copy already on `PATH` can't answer —
-     `uvx agents-shipgate@1.0.0a1 verify --preview --json` (or
-     `pipx run agents-shipgate==1.0.0a1 ...`). Only fall back to a bare
+     `uvx agents-shipgate@0.14.0 verify --preview --json` (or
+     `pipx run agents-shipgate==0.14.0 ...`). Only fall back to a bare
      `agents-shipgate verify --preview --json` once `agents-shipgate --version`
      confirms contract v9 or newer. Then follow
      [`prompts/add-shipgate-to-repo.md`](https://github.com/ThreeMoonsLab/agents-shipgate/blob/main/prompts/add-shipgate-to-repo.md)
      for the first-adoption helper flow, or point the user at the GitHub
-     Action (`ThreeMoonsLab/agents-shipgate@v1.0.0a1`) if they prefer CI.
+     Action (`ThreeMoonsLab/agents-shipgate@v0.14.0`) if they prefer CI.
    - If `run_shipgate: false` and `dry_run_recommended: true`: propose
      a non-mutating scan only — never propose `init --write` based on a
      dry-run match alone. Phrase it as "X may have shifted the tool

adoption-kits/claude-code-skill/prompts/stabilize-strict-mode.md

@@ -1,3 +1,3 @@
 # Prompt · Stabilize Agents Shipgate strict mode
 
 The user has Agents Shipgate running in **advisory** mode and wants to graduate to **strict** mode (CI fails on findings) without surprising contributors.
@@ -37,9 +37,9 @@
 
 5. **Update the CI workflow.** Replace the existing advisory step with strict + baseline. Use [`examples/github-actions/03-strict-with-baseline.yml`](https://github.com/ThreeMoonsLab/agents-shipgate/blob/main/examples/github-actions/03-strict-with-baseline.yml) as the template:
    ```yaml
-   - uses: ThreeMoonsLab/agents-shipgate@v1.0.0a1
+   - uses: ThreeMoonsLab/agents-shipgate@v0.14.0
      with:
-       shipgate_version: '1.0.0a1'
+       shipgate_version: '0.14.0'
        ci_mode: strict
        fail_on: critical
        baseline: .agents-shipgate/baseline.json

adoption-kits/codex-skill/assets/advisory-pr-comment.yml

@@ -18,9 +18,9 @@ jobs:
       - uses: actions/checkout@v4
         with:
           fetch-depth: 0
-      - uses: ThreeMoonsLab/agents-shipgate@v1.0.0a1
+      - uses: ThreeMoonsLab/agents-shipgate@v0.14.0
         with:
           ci_mode: advisory
           diff_base: target
           pr_comment: 'true'
-          shipgate_version: '1.0.0a1'
+          shipgate_version: '0.14.0'

docs/agent-contract-current.md

@@ -30,7 +30,7 @@ Downstream repos generated with
 `init --agent-instructions=default` get the minimal local copy at
 `.shipgate/agent-contract.json`.
 
-- Latest release: `v1.0.0a1` (see [pyproject.toml](../pyproject.toml) for the in-tree version)
+- Latest release: `v0.14.0` (see [pyproject.toml](../pyproject.toml) for the in-tree version)
 - Runtime contract: `9`
 - Current report schema: `0.28` — [`docs/report-schema.v0.28.json`](report-schema.v0.28.json)
 - Current packet schema: `0.7` — [`docs/packet-schema.v0.7.json`](packet-schema.v0.7.json)
@@ -362,7 +362,7 @@ exactly one stdout JSON object using
 `schema_version: "shipgate.codex_boundary_result/v1"` and the schema in
 [`codex-boundary-result-schema.v1.json`](codex-boundary-result-schema.v1.json).
 The removed `--format agent-json` alias and `agent_result_v1` schema string are
-breaking 1.0.0-alpha changes; see [STABILITY.md](../STABILITY.md#migration-note-100-alpha).
+breaking 0.14.0 changes; see [STABILITY.md](../STABILITY.md#migration-note-0-14-0).
 
 Coding agents should switch on `decision`, `completion_allowed`, `must_stop`,
 `first_next_action`, `human_review`, `repair`, and `policy`. Do not derive an agent

docs/agent-handoff-schema.v1.json

@@ -430,7 +430,7 @@
           "type": "string"
         },
         "version": {
-          "default": "1.0.0a1",
+          "default": "0.14.0",
           "title": "Version",
           "type": "string"
         }

docs/agent-result-schema.v1.json

@@ -369,7 +369,7 @@
           "type": "string"
         },
         "version": {
-          "default": "1.0.0a1",
+          "default": "0.14.0",
           "title": "Version",
           "type": "string"
         }

docs/agents/protocol.md

@@ -63,7 +63,7 @@ The stdout object has:
 
 Consumers must make decisions from JSON fields, never from prose or Markdown.
 The stable schema is `docs/codex-boundary-result-schema.v1.json`. The
-`1.0.0-alpha` contract renamed this local boundary result away from the older
+`0.14.0` contract renamed this local boundary result away from the older
 generic `agent_result_v1` schema string. `decision`, `completion_allowed`, `must_stop`,
 `first_next_action`, `human_review`, `repair`, and `policy` are the control
 signals. `risk_level` is explanatory and may differ between local-check and

docs/codex-boundary-result-schema.v1.json

@@ -369,7 +369,7 @@
           "type": "string"
         },
         "version": {
-          "default": "1.0.0a1",
+          "default": "0.14.0",
           "title": "Version",
           "type": "string"
         }